Cetuximab-Activated Natural Killer and Dendritic Cells Collaborate to Trigger Tumor Antigen–Specific T-cell Immunity in Head and Neck Cancer Patients

Tumor antigen–specific monoclonal antibodies (mAb) are clinically effective in a variety of malignancies. However, these results are only seen in a subset (∼20%) of patients (1). The large majority of treated patients, who are not likely to benefit from this therapy, emphasizes the need to characterize the mechanisms underlying the activity of tumor antigen–specific mAb and to optimize the selection of patients to be treated. In the present study, we have used cetuximab, an EGF receptor (EGFR)-specific, chimeric immunoglobulin G (IgG)1 isotype mAb that is U.S. Food and Drug Administration (FDA)-approved for colorectal carcinoma (CRC) and head and neck cancer (HNC), as a model. Despite overexpression of EGFR in 80% to 90% of HNC, cetuximab is effective in only in 10% to 20% of patients (2, 3).

A growing body of experimental and clinical evidence suggests that mAb-induced antitumor immunity underlie treated patients' clinical responses and may play a role in the differential clinical responses. Specifically, the statistically significant correlations of mAb-binding Fcγ receptor (FcγR) polymorphisms with clinical outcome in patients treated with rituximab (4, 5) trastuzumab (6, 7), and cetuximab (8, 9) argue in favor of a role of immune cell activation in the clinical responses, which have been described. Recently, trastuzumab emtansine (T-DM1) an antibody–drug conjugate showed significantly prolonged progression-free and overall survival in patients with breast cancer (10), and such a strategy may be enhanced by immune cell activation. However, the nonabsolute nature of the correlations in these malignancies (8, 9), the strong influence of k-ras mutation (8) on clinical response to cetuximab in CRC, and conflicting information about the beneficial effect of FcγRIIIa genotype in patients with CRC (8, 9), raise the possibility of additional immunologic mechanisms in the clinical benefit from mAb-based immunotherapy. This possibility is supported by the growing evidence in animal model systems (11) and in clinical settings (12) that tumor antigen–specific mAb may elicit cellular immunity.

Therefore, in this study, we have investigated whether the FcγRIIIa polymorphism on natural killer (NK) cells correlates with the antitumor activity of cetuximab using human HNC cells and with the clinical course of the disease in patients with HNC. We have selected these patients for our investigations, as in recent studies, we and others have shown in in vitro experiments a significant correlation of FcγRIIIa polymorphism with the antitumor activity of cetuximab (13). Furthermore, we have tested whether the interaction of cetuximab with FcγRIIIa on NK cells was required to trigger dendritic cell (DC) maturation and tumor antigen–specific cellular immune responses in patients with HNC. We show for the first time that cetuximab-activated NK cells trigger cross-talk and maturation of DC in an FcγR and NK group 2, member D (NKG2D)-dependent manner, and this results in tumor antigen–specific priming of CTL in cetuximab-treated patients with HNC. Finally, we have analyzed the mechanism(s) underlying the tumor antigen–specific immune response elicited by cetuximab and its potential clinical relevance.

The HNC cell lines HLA-A2⁻EGFR⁺ PCI-15B, HLA-A2⁻EGFR⁺, and MAGE-3⁺-JHU-029 (14–16), the breast cancer cell line MCF-7, and the lymphoid T2 cell line were grown in Iscove's modified Dulbecco's medium (IMDM; Sigma) supplemented with 10% FBS (Cellgro), 2% l-glutamine, and 1% penicillin/streptomycin (Invitrogen) at 37°C in a 5% CO₂, 95% humidity. Adherent tumor cells were detached by warm Trypsin–EDTA (0.25%) solution (Invitrogen).

The cohort of 107 cetuximab-treated patients with stage III/IV HNC described in Fig. 1 combined 60 patients enrolled on 2 prospective, cetuximab containing clinical trial regimens, UPCI-05-003 and UPCI-05-005 and 47 additional patients treated with cetuximab off protocol, as described in Table 1. The majority of these patients were treated with cetuximab plus cisplatin/paclitaxel/radiotherapy (UPCI-05-003; ref. 17) or cetuximab plus pemetrexed/radiotherapy (UPCI-05-005; ref. 18). Both trial cohorts were single arm phase II trials for locoregionally advanced, previously untreated disease. Patients were assigned to either trial by the treating physician at the time. The remainder of the patients was treated off-trial with cetuximab alone or in conjunction with palliative radiotherapy. EGFR tetramer measurements were conducted on protocol patients who were receiving single-agent cetuximab during the 6-month cetuximab maintenance phase of UPCI-05-003 (Table 1) or other newly diagnosed patients with HNC with stage III–IV disease while receiving cetuximab alone as primary treatment on a newly initiated, prospective phase II trial of single-agent cetuximab (UPCI-08-013). The comparison (cetuximab-naïve) HNC cohorts were gender and age-matched, previously cetuximab-untreated patients with HNC. No patients were excluded as a result of prior treatments or performance status. Blood from cetuximab-naïve patients with HNC was drawn within the same period after completing therapy without cetuximab.

The EGFR-specific chimeric IgG1 mAb cetuximab (Erbitux; BMS Imclone) and the EGFR-specific human IgG2 mAb panitumumab (Vectibix; Amgen) were purchased from the University of Pittsburgh Hillman Cancer Center Pharmacy (Pittsburgh, PA). The nonspecific control, human isotype IgG1 was purchased from Axxora LLC. Fluorescein isothiocyanate (FITC) or PE-Cy7-anti-CD11c mAb (R&D Systems), anti-CD80 mAb, anti-CD86 mAb, anti-CD83 mAb, anti-HLA-DR mAb, anti-CCR7 mAb (BD Biosciences Pharmingen) were purchased. Neutralizing anti-Fc FcγRIIIa antibody and phycoerythrin (PE)- and FITC-conjugated IgG isotypes for flow cytometry were purchased from BD Biosciences. FITC-goat anti-human Fc-specific IgG and FITC-goat anti-mouse IgG were purchased from Invitrogen. The antigen processing machinery (APM) components TAP-1–specific mAb NOB1 and TAP-2–specific mAb NOB2 were developed and characterized as described (19). Anti-CXCL10 antibody, recombinant human granulocyte macrophage colony-stimulating factor (GM-CSF), rec. human IFN-α, rec. human interleukin (IL)-1β, rec. human IL-2, rec. human IL-4, rec. human IL-6, rec. human IL-7, PGE2, poly I:C, and TNF-α were purchased from R&D Systems Inc.. IFN-γ was purchased from InterMune.

After approval by our Institutional Review Board [University of Pittsburgh Cancer Institute (UPCI; Pittsburgh, PA) protocol 99-069], informed consent was obtained from each subject. Blood from healthy donors (Western PA blood bank) or patients with HNC treated with cetuximab during or within 1 month of treatment. Lymphocytes were purified by Ficoll-Paque PLUS centrifugation (Amersham Biosciences) and stored frozen. DC were generated as described previously (19). NK cells and CD8⁺ T cells were purified using EasySep kits (Stem cell technologies) and purity was more than 95% FcγRIIIa⁺, CD56⁺, and CD3⁻ (13).

Peripheral blood mononuclear cell (PBMC) were genotyped at FcγRIIIa (codon 158) using quantitative PCR (Applied Biosystems). Genomic DNA was extracted using the DNeasy Kit (Qiagen) following the manufacturer's protocol. Genomic DNA (5–50 ng) was added to a 25-μL reaction using 2× TaqMan master mix (Applied Biosystems). Plates were analyzed using an ABI prism 7700 (13).

Mature DC were incubated with autologous negatively isolated CD8⁺ T cells for 48 hours at 37°C with rhIL-2 (20 U/mL) and rhIL-7 (5 ng/mL). On day 7 and weekly thereafter, lymphocytes were restimulated with autologous irradiated DC pulsed with peptide in culture medium supplemented with IL-2 (20 U/mL) and IL-7 (5 ng/mL). The stimulated CD8⁺ T cells were analyzed for tumor antigen specificity in IFN-γ ELISPOT at day 21 and then every 7 days thereafter. We have previously published a novel wild-type EFGR_853–861–specific tetramer that recognizes EFGR_853–861–specific CTL and also showed that this specific tetramer recognizes EFGR_853–861–specific CTL clone that was generated by pulsing DC with EFGR_853–861 peptide in vitro (14).

IFN-γ ELISPOT assays were conducted as described previously (14). The MAGE-3_271–279 (FLWGPRALV) peptide and wild-type EGFR_853–861 peptide (ITDFGLAKL) were produced by the peptide synthesis facility at the University of Pittsburgh (Pittsburgh, PA), using F-moc technology (19, 20). The HLA-A2-MAGE-3_271–279 tetramer was synthesized by the NIH Tetramer Facility (Emory University, Atlanta, GA). CTL lines were sorted using PE-labeled HLA-A2-EGFR_853–861 peptide-loaded tetramers (14) and restimulated every 7 to 10 days. For ELISPOT assays, CD8⁺ T cells were seeded in triplicate (5 × 10⁴) for bulk CD8⁺ T cells and 1 × 10³ for CD8⁺ T-cell clones) in multiscreen HTS plates. Synthetic peptides were then added to ELISPOT assays after APCs were seeded. After 18 hours, spots were enumerated using computer-assisted image analysis software.

Lymphocytes were prepared for flow cytometry by washing with PBS (Sigma-Aldrich) and fluorescence-activated cell sorting (FACS0 buffer (2% FBS in PBS). For flow-cytometric analysis using HLA-A2-peptide tetramers, PE-labeled HLA-A2-EGFR_853–861 tetramers were obtained from the Tetramer Facility of the NIH (Atlanta, GA). Cells were analyzed on cyan flow cytometer (Beckman Coulter) using Summit v4.3 software. Intracellular IFN-γ staining of CD56⁺ NK cells was conducted by using cytofix/cytoperm fixation/permeabilization kit (BD Biosciences). Briefly, cells were treated with BD GolgiSTOP reagent and stained with FITC-anti-CD56 antibody. After 4 hours, cells were washed, fixed, permeabilized, and stained with FITC-anti-IFN-γ antibody.

A standard calibration curve was generated by serial dilutions of recombinant cytokine. Cytokine concentrations in culture supernatants were determined by a multiplexed ELISA (Luminex) at the UPCI facility. Hu30 Plex kit (Invitrogen, catalog number LHC 6003) detected IL-1β, IL-1RA, IL-2, IL-2R, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 (p40/p70), IL-13, IL-15, IL-17, TNF-α, IFN-α, IFN-γ, GM-CSF, MIP-1α, MCP-1,MIP-1β, CXCL-10, MIG, Eotaxin, RANTES, VEGF, granulocyte colony-stimulating factor (G-CSF), EGF, basic fibroblast growth factor (bFGF), and hepatocyte growth factor (HGF).

Cell migration assays was conducted in 5.0-μm pore size 96-well polycarbonate filtration plate (Millipore). Briefly, the lower chambers were filled with cell-free culture supernatant collected from various experimental conditions. T cells (5 × 10⁴/100 μl of fresh media) were plated in the upper well. Following 4-hour incubation at 37°C, the upper wells containing filtration unit were gently removed and the cells in bottom chamber were counted by hemocytometer. After 4 hours incubation and counting, cell numbers were plotted as the percentage migrated cells.

Disease-specific survival (DSS) of patients with HNC was defined as the time elapsed from the first treatment with cetuximab until death from HNC. Patients were censored if they were alive at last follow-up or had died but were cancer-free at the time of death. FcγR genotype subclasses were tested for association with disease-free survival with the log-rank test. Other factors were tested for association with FcγR subclasses including cancer type (primary vs. recurrent), disease site and whether the patient was treated on protocol. These results were used to insure the conclusions about the influence of FcγR upon disease-specific mortality were not confounded by other potentially influential covariates. Two-tailed unpaired t test was conducted for statistical analysis to compare the significant difference between 2 groups and P < 0.05 was considered significant. T-cell reactivity as measured by the ELISPOT assay was considered positive if the number of spots in test wells was significantly higher than that in background wells when using a one-tailed permutation test for α ≤ 0.05.

We have previously shown that NK cells expressing an FcγRIIIa V–encoding allele (at codon158) are more potent than NK cells expressing an FcγRIIIa F–encoding allele in antibody-dependent cell-mediated cytotoxicity (ADCC) against HNC cells in vitro (13). The clinical significance of our previously published in vitro data (13, 15) was then investigated in a cohort of 107 consecutive patients with HNC treated with cetuximab at our institution and genotyped for polymorphic FcγRIIIa position 158 VV, VF, or FF. Interestingly, FcγRIIIa genotype was not associated with DSS in this large cohort of cetuximab-treated patients with HNC (Fig. 1 and Table 1). The low frequency of the beneficial VV genotype (5%; ref. 13), in light of the higher observed rate of clinical response (∼15%–20%; ref. 21), raised the likelihood that an additional immune mechanism may play a role in the antitumor activity of cetuximab. Because cetuximab can activate NK cells in the presence of EGFR⁺ tumor cells, and subsequently NK cells may provide a crucial DC maturation signal, we hypothesized that cetuximab-activated NK cells might facilitate the generation of tumor antigen-specific CTL.

To evaluate whether induction of CD8⁺ T cells by cetuximab occurred in vivo, we measured the frequency of EGFR_853–861–specific CTL in the circulation of cetuximab-treated patients with HNC. PBMC were obtained from HLA-A2⁺ cetuximab-treated (n = 17) and HLA-A2⁺ cetuximab-naïve patients with HNC (n = 39), and compared with HLA-A2⁺ healthy donors (n = 24). On the basis of the analysis of CD3⁺CD8⁺ T cells stained with an EGFR_853–861–specific tetramer (14), a significantly higher frequency of EGFR_853–861–specific T cells was found in cetuximab-treated patients with HNC than in cetuximab-naïve patients with HNC (Fig. 2A). As a control, we observed low background of EGFR_853–861–specific precursor CTL in healthy HLA-A2⁺ donors. These data strongly support that the administration of cetuximab to patients with HNC leads to EGFR_853–861 peptide cross-presentation by DC, triggering expansion of EGFR-specific T cells. Similarly, higher frequency of EGFR_853–861–specific T cells was also found in cetuximab-treated patients with HNC who received 4 weeks of single-agent cetuximab treatment (Fig. 2B and C). Taken together, these results indicate that cetuximab alone or in combination with chemotherapy induces expansion of EGFR-specific T cells.

Having observed a higher level of EGFR-specific T cells in cetuximab-treated patients with HNC, we investigated in vitro whether cetuximab enhanced EGFR-specific T-cell immunity and whether this effect was modulated by NK cells. We studied the effect of cetuximab with or without NK cells on DC cross-presentation using EGFR⁺, HLA-A2⁻ HNC cells to stimulate cognate CTL (19). First, DC alone (HLA-A2⁺ donors) were incubated with cetuximab and HLA-A2⁻ PCI-15B HNC cells, which did not show enhanced cross-presentation. We furthermore included FcγRIIIa⁺ NK cells with DC to evaluate their effect on the cross-presentation capacity of DC. Although addition of NK cells alone showed a 3-fold increase in the cross-presentation by DC, the addition of cetuximab + NK cells resulted in a 5-fold enhancement in the EGFR_853–861–specific CTL activation in comparison with IgG1 control mAb + NK cells (P < 0.001). The cross-presentation by NK cell–stimulated, HLA-A2⁺ DC was quantified using IFN-γ ELISPOT assays and HLA-A2⁺, EGFR_853–861–specific CTL (14). DC incubated with HLA-A2⁻ PCI-15B HNC cells and cetuximab enhanced NK cell–dependent cross-presentation to EGFR-specific CTL. A low basal level of cross-presentation was observed with apoptotic PCI-15B HNC cells incubated with an isotype control IgG1 or without NK cells (Fig. 2D) or without HNC cells. A positive control included CTL recognition of EGFR_853–861 peptide loaded DC.

We selected another HNC cell line JHU-029, which expresses EGFR as well as MAGE-3 (14, 20), to investigate whether cross-presentation of MAGE-3 by DC could also be enhanced by cetuximab-activated NK cells. DC incubated with HLA-A2⁻ JHU-029 HNC cells and cetuximab significantly enhanced the cross-presentation to MAGE-3_271–279 peptide–specific CTL, primarily in the presence of NK cells. Addition of NK cells alone showed a 2-fold increase in EGFR cross-presentation by DC. Moreover, addition of cetuximab with NK cells resulted in approximately 4-fold enhancement in the MAGE-3_271–279 peptide–specific CTL activation in comparison with control IgG1 mAb (P < 0.01; Fig. 2E). Cytokine-matured DC alone were used as a control in the cross-presentation assay. An isotype control IgG1 failed to facilitate enhancement of cross-presentation of EGFR or MAGE-3 antigens in the presence or absence of NK cells. The importance of FcγRIIIa-activated NK cells to promote the activation of EGFR-specific (Fig. 2D) as well as non-EGFR antigen (MAGE-3)-specific CD8⁺ T cells, highlights the beneficial effect of enhanced DC maturation and antigen presentation after NK cell activation by cetuximab (Fig. 2D and E).

To model cetuximab-dependent cross-priming of EGFR to CTL, we conducted in vitro stimulation for 2 weeks, with DC matured using NK cells activated by PCI-15B HNC cells coated with cetuximab. Consequently, we measured the efficiency of cross-priming of EGFR-specific CTL by tetramer quantification by flow cytometry. DC matured with cetuximab-activated NK cells and PCI-15B HNC cells induced significantly higher levels of EGFR-specific CTL than panitumumab. The latter IgG2 mAb was used as control mAb (Supplementary Fig. S1) due to its reduced FcγRIIIa binding. Taken together, these data strongly support that the administration of cetuximab to patients with HNC leads to EGFR_853–861 peptide cross-presentation by DC, triggering expansion of EGFR-specific T cells.

To determine whether enhanced tumor antigen cross-presentation by cetuximab was mediated through enhanced DC maturation (19), we analyzed the expression level of surface maturation markers by DC incubated with cetuximab, PCI-15B HNC cells and NK cells. FACS analysis of DC showed significant upregulation of HLA-DR, costimulatory molecule CD80, CD83, and CD86 (Fig. 3A and Supplementary Fig. S2), and APM component TAP-1 (Fig. 3B). The latter is highly associated with efficient DC cross-presentation (19). Similarly, increased HLA-DR⁺/CD86⁺ expression was observed on the cetuximab-activated, NK cell–treated DC (Fig. 3C). Enhanced frequencies of HLA-DR⁺CD83⁺ DC or HLA-DR⁺/CD86⁺ DC were also observed after culture with cetuximab-activated NK cells (Supplementary Fig. S3A and S3B). Interestingly, enhanced expression of CD137 by DC was also observed after culture with cetuximab-activated NK cells (Fig. 3D and E; ref. 22).

Although differences in lytic activity between NK cells expressing the FcγRIIIa VV genotype versus those expressing the FF genotype were reported (13), both VV and FF expressing, cetuximab-activated NK cells showed a similar effect on the expression of the DC maturation markers, CD80 and TAP-1 (Unpublished Data). Despite the lack of influence of codon 158 polymorphism on NK cell-induced DC maturation, the absolute dependence of FcγRIIIa binding by NK cells was evident, as the EGFR-specific mAb panitumumab, which binds very poorly to FcγRIIIa due to its IgG2 isotype (23), did not mediate downstream DC maturation (Fig. 3B and Supplementary Fig. S2) or NK cell activation (Fig. 4A).

Next, we investigated the cytokine(s) secreted by cetuximab-activated helper NK cells, which could be responsible for this DC maturation. Cetuximab treatment of NK cells in the presence of PCI-15B HNC cells was found to induce a high level of IFN-γ secretion by NK cells (Fig. 4A). Moreover, cisplatin does not negatively affect cetuximab-induced activation and secretion of IFN-γ by NK cells (Fig. 4A). Notably, cetuximab-activated, NK cell–treated DC further stimulated these NK cells in a reciprocal fashion, leading to significantly increased secretion of IFN-γ by cetuximab-activated NK cells when incubated with DC. Control IgG1 did not activate NK cells (Fig. 4B), consistent with the need for mAb binding of both EGFR and FcγRIIIa. Furthermore, the absence of cetuximab in the NK:DC coculture (without HNC cells) or DC:PCI-15B coculture (without NK cells) abrogated the IFN-γ secretion (Supplementary Figs. S4 and S5). We then determined that DC maturation was mediated through IFN-γ, as an IFN-γ-specific neutralizing antibody blocked the induction of HLA-DR (Fig. 4C) on DC by cetuximab-activated NK cells. Blockade of other cytokines, such as IL-12p40/70 showed a slight ability to promote HLA-DR expression on DC, however, no significant differences were found on DC maturation (Fig. 4C). Similarly, a notable effect of the blocking of IFN-γ was observed on the maturation markers HLA-DR, CD80, HLA-DR⁺/CD86⁺, or HLA-DR/CD83⁺ on DC by cetuximab-activated NK cells (Supplementary Fig. S3A and S3B). These changes in the expression of DC maturation markers are specific, as no significant increase in their expression level was detected on DC after incubation with autologous NK cells (Supplementary Fig. S5), PCI-15B cells alone or isotype control IgG1 (Supplementary Figs. S2 and S3).

In the experiments described earlier (Fig. 2D), we observed that DC cross-priming of EGFR-specific T cells was enhanced in the presence of cetuximab-activated NK cells. The latter cells had been previously shown to secrete high levels of T_H1 cytokines and chemokines (13, 24). Because these cytokines strongly upregulate the expression of certain APM components, such as TAP-1 and TAP-2, which are crucial for tumor antigen–derived peptide presentation to cognate CTL (19), we investigated whether these APM components were upregulated in DC incubated with cetuximab-activated NK cells. Consistent with this notion, the enhanced level of TAP-1 was observed in DC that were cocultured with cetuximab-activated NK cells, even if they were physically separated from the NK cells:PCI-15B cells coculture (in the presence of cetuximab) through a Transwell membrane (Fig. 4D). The increase in TAP-1 expression in cetuximab-activated, NK cell–treated DC was found after incubation with PCI-15B HNC cells (EGFR^high) but not with MCF-7 breast cancer cells. These results indicate that high EGFR expression on tumor cells is necessary for cetuximab-induced NK cell activation to induce DC maturation (Fig. 4E). Also, the upregulation of TAP-1 in DC cocultured with cetuximab-activated NK cells was abrogated by the blocking FcγRIIIa–specific mAb 3G8 but was not affected by a control isotype IgG1 (Fig. 4F). Furthermore, incubation of NK cells with the EGFR-specific, IgG2 isotype mAb panitumumab which binds poorly to FcγRIIIa on NK cells (15, 23, 25), failed to upregulate the expression of costimulatory molecules on DC, in accordance with the lack of effect on EGFR cross-priming (Supplementary Figs. S1–S3). Thus, cetuximab-bound HNC cells activate NK cells through an FcγRIIIa-dependent mechanism, leading to APM component and costimulatory molecule upregulation in DC, manifested by their enhanced ability to mediate tumor antigen cross-presentation.

In light of the reciprocal activation of NK cells by DC (Fig. 4B), we investigated the molecular mechanism underlying the interaction of cetuximab-activated NK cells cultured with mature DC. Specifically blocking mAb for known modulators of NK–DC cross-talk were used to disrupt NK:DC interactions. Interestingly, only a blocking NKG2D-specific mAb abrogated the reciprocal activation of cetuximab-activated NK cells and DC, leading to significantly reduced IFN-γ secretion (Fig. 5A). Blockade of other important contributors to NK–DC cross-talk, including NKp30 expressed by NK cells, as well as IL-12 and IL-18, were not responsible for this effect. We observed upregulation of MHC class I chain-related protein A (MICA) on DC but not on HNC cells, suggesting a major role of NKG2D in the reciprocal NK cell activation mediated by cetuximab-induced DC maturation (Fig. 5B).

The consequences of NK cell–DC stimulation include functional and phenotypic activation, such as secretion of chemokines and cytokines. We conducted multiplex ELISA (Luminex) analysis to profile the cytokines and chemokines in the supernatants of coculture (DC:PCI-15B or DC:NK:PCI-15B) untreated or treated with IgG1 or cetuximab. Secretion of the cytokines and chemokines, MCP-1, MIP-1β, IL-12p40/70, (CCR5 ligand), CXCL10, and MIG (CXCR3 ligands) was enhanced only in the cetuximab-activated NK cell–treated, DC (Figs. 6A–E). Because cetuximab-activated, NK cell–treated DC supernatants contained higher amounts of chemokines responsible for recruiting and activating CD8⁺ T cells, we confirmed the enhanced migratory ability of naïve CD8⁺ T cells under the influence of chemotactic factor/s present in supernatant from cetuximab-activated NK cell–treated DC. The induced T-cell migration was completely abrogated by a CXCL10-specific mAb (Fig. 6F).

While blockade of proliferative and antiapoptotic pathways in tumor cells plays a major role in the clinical efficacy of therapeutic mAb, such as rituximab, cetuximab, and trastuzumab, the role of FcγRIIIa–dependent innate immunity (ADCC) induced by these antitumor mAbs have generated conflicting findings (8, 9, 26–29). In addition, using murine models, the role of adaptive antitumor immune mechanisms in the clinical activity of these mAb has been suggested (30, 31), though data from treated patients with cancer is generally lacking. Our in vitro data showed that NK cells significantly enhance the antitumor effects of cetuximab via interaction with FcγRIIIa (13), which parallel similar results obtained with trastuzumab in breast carcinoma bearing mice (7, 32). Taken together, these results indicated that FcγRIIIa on NK cells could play a critical role in an immuno-naïve environment. However, this effect becomes less marked in “immunocompetent” patients with cancer, as we did not find a significant correlation between FcγRIIIa genotype and disease-free survival in patients with HNC treated with cetuximab and radiotherapy and chemotherapy (Fig. 1), as seen in patients with other cancer (26–29). Nevertheless, the data are in agreement with the dominant role of FcγRIIIa on NK cells, although not absolutely dependent on the index of affinity of cetuximab, but rather to initiate proximity of FcγR bearing NK cells and DC with EGFR⁺ HNC cells and with each other, leading to NK–DC cross-talk mediated through NKG2D-MICA. NK–DC cross-talk is crucial to cetuximab-mediated DC maturation and cross-presentation, which may potentially lead to CTL expansion, and potentially clinical effects, in vivo in cetuximab-treated patients with HNC. While the sample size of this retrospective analysis was insufficient to correlate T-cell frequencies or phenotype with clinical outcome, such a study is ongoing as part of a separate, prospectively treated cohort. Nevertheless, higher frequency of EGFR-specific T cells was also found in prospectively treated patients with HNC who received cetuximab alone (without chemotherapy), strongly supporting our conclusions (Fig. 2B and C).

Interestingly, that adaptive immune responses might contribute to therapeutic efficacy, based on higher frequency of EGFR-specific CD8⁺ T cells, must be validated prospectively in a larger cohort cetuximab-treated patients (Fig. 2) with sufficient follow up (2–3 years) for outcome measurements. This extends recent reports of rituximab and anti-HER2/neu inducing long lasting antitumor adaptive immune responses in murine systems, as support the plausible role of adaptive antitumor immune responses triggered by antitumor specific mAb in treated patients (30, 31, 33). The dependence on IFN-γ secretion by NK cells supported that NK cells provide helper function in DC cross-presentation, extending our previous finding that an immunogenic EGFR-encoded CTL could be generated by pulsing DC with EGFR_853–861 peptide that could stimulate cognate CTL in vitro (14). However, total recombinant EGFR protein was not used in this study as a source of tumor antigen as this would not mimic FcγRIIIa–mediated effects and NK:DC cross-talk that cetuximab stimulates.

Although FcγRIIIa polymorphism-dependent survival of patients with cancer were reported in CRC cohorts (8, 9), adaptive immune responses probably contribute in the cetuximab-mediated immune responses and may better predict clinical outcome of immunotherapy, because NK cells also provide helper function in shaping host adaptive immune responses. Moreover, the low frequency of VV patients often “negatively” influences statistical analysis (26, 29). These data clearly support the paradigm that antitumor mAb participate in the long-term control of tumor in patients with HNC, probably accounting for the clinical activity (30, 31, 33–35).

To analyze the importance of FcγRIIIa, we used the 2 FDA-approved, EGFR-specific mAb cetuximab and panitumumab, which bind the same epitope on EGFR (23, 25) but differ in their isotype and FcγRIIIa affinity, to show that only cetuximab (IgG1)–induced NK–DC cross-talk and DC cross-presentation ability, whereas panitumumab (IgG2) does not, despite their equal EGFR binding (Fig. 3B and Supplementary Figs. S1 and S2). It is notable that this IgG2 isotype mAb, despite shown potency in EGFR signaling blockade, has failed multiple prospective HNC trials of clinical efficacy, supporting the proposition that cetuximab imparts greater clinical activity due to the IgG1 isotype-induced immunologic effects.

We have recently shown that cetuximab-naïve patients with HNC possess elevated levels of EGFR_853–861–specific T cells confirming that EGFR expressed on HNC cells induces a specific immune response in vivo (36). Although antigen-specific CD4⁺ T-cell activation has been noted in a high proportion of patients with breast cancer treated with trastuzumab and correlated with clinical responses (7), this is the first demonstration of enhanced tumor antigen–specific CTL in mAb-treated patients (Fig. 2), which are crucial to mediating strong antitumor immunity in vivo. This effect provides a general yet novel immune mechanism of enhanced antitumor activity relevant to clinical response, as well as a potential biomarker to monitor cetuximab effects in mAb-treated patients with cancer. Because the frequency of HLA-A2.1 in patients with HNC in different regions varies (ranging from 30%–50% worldwide), additional HLA-restricted, EGFR-specific T-cell epitopes should be identified to expand application of such a biomarker.

Our results indicate that cetuximab:EGFR⁺ tumor cell complexes originating in the presence of NK cells are provided to DC to generate polyclonal tumor antigen (MAGE-3 as well as EGFR) processing and presentation (36–38). In addition, activated-NK cells facilitate DC maturation, as indicated by the concomitant upregulation of the costimulatory molecules CD80, CD86, and CD137 on the DC surface (Fig. 3A–E and Supplementary Figs. S2 and S3). At the same time, cytokines released by NK cells activated through binding of EGFR-cetuximab complexes via FcγRIIIa induce activation markers and TAP-1 (refs. 39–42; Figs. 3B and 4A–F), which we have shown to predict cross-priming capacity of DC (19). Our data show that with the release of cytokines from cetuximab-activated NK cells DC becomes highly responsive to the tumor microenvironment and undergoes maturation along with uptake of dead tumor materials (37). This finding is in strong accordance with the established fact (24, 37) that release of cytokines from cetuximab-activated NK cells may trigger T_H1 responses against tumor microenvironment. The ability of activated/mature DC to generate or activate EGFR-specific CTL is an attractive opportunity to subvert immunosuppressive environment, which could be easily achieved with antitumor mAb administrations (43). Through this mechanism, therapeutic mAb may enhance antigen cross-presentation by DC to T cells, resulting in augmentation of tumor antigen–specific CTL generation (43–45). If this interpretation is correct, cetuximab administration should be probably combined with vaccines targeting EGFR (14) or other tumor antigen, as well as adjuvant(s) to specifically stimulate cellular immunity.

In addition to their ability to mediate ADCC, cetuximab-activated NK cells, secrete cytokines, such as IFN-γ, MCP-1, MIP-1β that inhibit tumor cell proliferation, enhance antigen presentation, and chemokines such as IP10 and MIG that aid in the chemotaxis of T cells (13, 39). The resulting potent activating bi-directional signaling between NK cells and DC can shape both the innate immune response within inflamed peripheral tissues and the adaptive immune response in secondary lymphoid organs (46) enhancing cross-presentation and priming of T cells, as we observed (Figs. 2 and 3). Through direct cellular interactions and secretion of cytokines/chemokines NK cells function as helper cells (40, 47, 48) and enhance DC cross-presentation and T-cell induction, with the potential to strengthen tumor antigen–specific cellular immune response and to spread this priming to multiple tumor antigen, including private antigens. The effects observed can be blocked using a neutralizing IFN-γ mAb consistent with our previous findings about optimal APM upregulation in mature DC mediating cross-presentation (19). However, the beginning of DC maturation is clearly dependent on secreted IFN-γ from NK cells, whereas IL-12 might contribute for the cross-priming activities and reciprocal NK cell activation (Figs. 4B and C and 5A and B). We have also observed that NKG2D, a potent tumor recognition molecule on NK cells may interplay in shaping the NK-DC cross-talk and reciprocal activities, subsequently NKG2D can also interact with its ligand expressed on tumor cells and or activated DC (Fig. 5).

However, the generation and functional activation of NK cells and CTL for elimination of tumor cells might be influenced by the impact of concomitant chemotherapy and/or radiotherapy (8, 22), often administered in conjunction with cetuximab (49). We observed that cetuximab-induced IFN-γ secretion by NK cells activation is not negatively influenced by cisplatin (Fig. 4A). Recently, trastuzumab emtansine (T-DM1) an antibody–drug conjugate showed a significantly prolonged progression-free and overall survival in patients with breast cancer (10). The induction of T cells might be more relevant and efficacy even further enhanced, when mAb-conjugated cytotoxics can induce immunogenic cell death. Currently, we are investigating the combination of cetuximab with NK cell–activating molecules and with cytotoxic drugs to further enhance the adaptive immune responses in patients and functional properties of HLA class I or APM components as potential immune escape mechanisms, especially in HNC cells (50). These mechanisms and variables are now being studied in prospective clinical trials using single-agent cetuximab, as well as cetuximab in combination with chemoradiotherapy, as in both situations clinical efficacy is observed.

No potential conflicts of interest were disclosed.

Conception and design: R.M. Srivastava, S.C. Lee, H.C. Davidson, A. López-Albaitero, S. Ferrone, R.L. Ferris

Development of methodology: S.C. Lee, C.A. Lord, H.C. Davidson, R.L. Ferris

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.M. Srivastava, S.C. Lee, P.A. Andrade Filho, C.A. Lord, H.-B. Jie, H.C. Davidson, A. López-Albaitero, S.P. Gibson, R.L. Ferris

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.M. Srivastava, S.C. Lee, C.A. Lord, H.-B. Jie, H.C. Davidson, W.E. Gooding, R.L. Ferris

Writing, review, and/or revision of the manuscript: R.M. Srivastava, P.A. Andrade Filho, C.A. Lord, H.C. Davidson, S. Ferrone, R.L. Ferris

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.P. Gibson, R.L. Ferris

Study supervision: R.L. Ferris

The authors thank Jennifer Ridge-Hetrick, who assisted in extracting data from the head and neck cancer database at the University of Pittsburgh Cancer Registry and the core University of Pittsburgh Cancer Institute flow cytometry and Luminex facilities. The authors also thank Drs. Ethan Argiris, Michael Gibson, and James Ohr for trial coordination and caring for cetuximab-treated patients studied here, and Ferris laboratory members for helpful suggestions.

This work was supported by NIH grants R01 DE19727, P50 CA097190, CA110249, and University of Pittsburgh Cancer Institute grant P30CA047904.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.